ARTICLE pubs.acs.org/IC
Exploratory Syntheses and Structures of SrAu4.3In1.7 and CaAg3.5In1.9: Electron-Poor Intermetallics with Diversified Polyanionic Frameworks That Are Derived from the CaAu4In2 Approximant Qisheng Lin and John D. Corbett* Department of Chemistry, Iowa State University, Ames, Iowa 50011, United States
bS Supporting Information ABSTRACT: The phase regions around quasicrystals and approximants (QC/ACs) are rich pools for electron-poor intermetallics with novel, complex structures, and bonding patterns. The present SrAu4.30(1)In1.70(1) (1) and CaAg3.54(1)In1.88(1) (2) were synthesized through chemical tunings of the model CaAu4In2 (YCd6Type) AC. Single crystal X-ray diffraction analyses reveals that crystal 1 has Pnma (CeCu6-type) symmetry, with a = 9.102(1) Å, b = 5.6379(9) Å, and c = 11.515(2) Å. The building block in 1 is a 19-vertex cluster Sr@Au9In4M6 (M = Au/In), which vividly mimics Ca@(Au,In)18 in Ca3Au12.4In6.1 (YCd6-type) in geometry. These clusters aggregate into one-dimensional columns extending along the b axis. Crystal 2 (P6/mmm, a = 20.660(3) Å, c = 9.410(2) Å) is closely related to Na26Cd141 (hP167) and Y13Pd40Sn31 (hP168), which are differentiated by the selective occupation of Wyckoff 1a (0 0 0) or 2d (1/3 2/3 1/2) sites by Cd or Pd. Crystal 2 adopts the Na26Cd141 structure, but the 1a site is split into two partially occupied sites. The synergistic disorder in the hexagonal tunnels along c is a major property. The valence electron count per atom (e/a) values for 1 and 2 are 1.63 and 1.74, respectively, the lowest among any other ternary phases in each system. These values are close to those of ACs in the CaAuM (M = Ga, In) systems. Electronic structures for both are discussed in terms of the results of TB-LMTO-ASA calculations.
’ INTRODUCTION Electron-poor intermetallics,1 by which we mean those phases that have low valence electron counts per atom (e/a), namely, close to or lower than 2.0, have recently attracted considerable attention because of their close relationship to the HumeRothery phases (e/a < 2.0). The last phases originally refer to distinctive structural types (brasses) in the CuZn binary system2,3 but recently the more structurally complex quasicrystals and approximant phases (1.72.2), abbreviated as QC/ACs, have also been included in this category.4 Electron-poor intermetallic phases are electronically positioned between the HumeRothery and the Zintl phases (e/a > 4.0).57 (Note that the last values have traditionally been calculated for the anions only.) Generally, the metalmetal bonding interactions among electronegative constituents in polar intermetallics exhibit mainly covalent character, similar to the HumeRothery phases, but the interactions between electropositive metals and polyanions may also show considerable covalent bonding depending on the extent of electron transfer between them. Electron-poor intermetallics have been poorly investigated, and the interplays among valence electron count, structure, and bonding are not well understood. Electron-poor polar intermetallics may be accessed by different routes. Besides decreases in the proportion of the alkali- or alkaline-earth metals, e/a values can also be decreased by alloying Zintl ions with electron-poorer transition metals,1 for example, group 11 metals, Au particularly. Another good place to start is r 2011 American Chemical Society
the phase regions around quasicrystals and approximants (QC/ACs).5 According to our earlier experimental results, such electron-poor polar intermetallics may have complex structures, as do QC/ACs, and unprecedented structural units and novel bonding patterns. For example, Sc4Mg0.5Cu14.5Ga7.6 is a polar intermetallic phase exhibiting an incommensurately modulated structure, and its chemical composition and e/a are very close to those of the Sc3Mg0.2Cu10.5Ga7.3 1/1 AC or Sc15Mg3Cu48Ga34 (= Sc3Mg0.6Cu9.6Ga6.8), the icosahedral QC.8 Other parallel electron-poor intermetallics discovered in such ways include Mg35Cu24Ga53,9 Ca3Au10In4,10 and Ca3Au8Ge3.11 During the past years, we have often utilized the combination of Ca and Au with other p-block elements in searches for QC/ ACs, for example, CaAuM (M = Ga,12,13 In,14 Ge,15 Sn16,17). In addition, we have also considered chemical tunings in terms of both the active metal Ca and the dominant Au. It appears that the active metals Ca, Yb, and Y play critical roles in the formation of the YCd6-type QC/ACs. On one hand, these metals have evidently suitable atomic sizes and electronegativities to match the electronegative components to meet the general conditions required by the HumeRothery rules.2,3 On the other hand, they have low-lying empty or partially filled d valence orbitals to interact effectively with s and p states of electronegative components, through which the depths of pseudogaps around Fermi energies are greatly enhanced. Received: July 28, 2011 Published: October 11, 2011 11091
dx.doi.org/10.1021/ic2016275 | Inorg. Chem. 2011, 50, 11091–11098
Inorganic Chemistry For this reason, Na and Mg are evidently not suitable for YCd6-type QC/ACs.14 The heavier alkali or alkaline-earth metals have larger sizes and also contain empty d orbitals, but no QC/AC containing K, Rb, Cs, Sr, and Ba has been reported. Systematic experimental studies are very much needed. Another feature is the influence of the magic Au (with the largest relativistic effects among all metals) in the formation of QC/ACs. Au appears to be unique as the major component (>50 at. %) in the formation of CaAuM QC/ACs (M = Ga,12,13 In,14 Ge,15 Sn16,17) in terms of both electronic and size effects. The question as to whether Au is a required constituent is also interesting. This work reports two different chemical tunings, SrAuxIn6x and CaAgyIn6y, derived from the earlier studies on CaAu4M2 (M = Ga, In, Ge, Sn)1316 systems to test whether Ca or Au is required. Not surprisingly, each results in a new but not a YCd6type phase, although the starting compositions and reaction conditions were very similar to those employed to gain such products.
ARTICLE
Table 1. Crystal Data and Structure Refinement for SrAu4.30(1)In1.70(1) (1) and CaAg3.54(1)In1.88(1) (2) formula f.w.
SrAu4.30(1)In1.70(1) 1130.18
Ca26.0(2)Ag92.0(1)In49.0(2)a 16592.3
space group/Z
Pnma/4
P6/mmm/1
a
9.102(1)
20.660(3)
b
5.6379(9)
20.660(3)
c
11.515(2)
9.410(2)
V (Å3)/dcal (g/cm3) abs. coeff. (mm1)
590.9(2)/12.70 121.77
3478.4(10)/7.92 21.48
refl. coll./Rint
3530/0.0474
22038/0.0414
data/restr./para.
778/0/42
1713/2/98
GOF on F2
1.245
1.14
R1/wR2 [I > 2 σ(I)]
0.0584/0.1487
0.0375/0.0775
[all data]
0.0615/0.1499
0.0386/0.0787
unit cell (Å)
’ EXPERIMENTAL SECTION
The formula, f.w., and dcal values for 2 are different from those in cif file (Supporting Information) in which Cd is used to represent Ag/In mixtures in the structural solution.
Syntheses. The starting materials were calcium granules (99.5%, Alfa-Aesar), dendritic strontium (99.9%, Alfa-Aesar), gold foil (Ames Lab, > 99.99%), silver shot (Ames Lab, > 99.99%), and indium ingots (99.99%, Alfa-Aesar) that were stored in a glovebox filled with nitrogen (H2O, < 1 ppmv). The surfaces of calcium, strontium, and indium metals were cleaned with a surgical blade before use. Mixtures were weld-sealed under an argon atmosphere into tantalum containers, which were later sealed within an evacuated SiO2 jackets (95% phase purity. Single-crystal X-ray diffraction data were collected at room temperature with the aid of a Bruker APEX CCD single crystal diffractometer equipped with graphite-monochromatized Mo Kα (λ = 0.71069 Å) radiation, and with exposure times of 30 s per frame. Data integration,
absorption, and Lorentz polarization corrections were made by the SAINT and SADABS subprograms in the SMART software packages.20 Assignments of the space groups from the diffractometer data were made on the basis of the Laue symmetry and systematic absence analyses. Structure refinements were performed with the aid of the SHELXTL subprogram.20 Direct methods for 1 yielded six atoms: five had separations suitable for Au/InAu/In pairs and the remaining one, for SrAu/In, so they were initially assigned to Au1Au5 and Sr, respectively. Subsequent least-squares refinements proceeded smoothly and R1 converged at 13.7%. Examination of the resulting isotropic displacement parameters revealed that the assigned Au4 and Au5 had larger Ueq values (0.031 and 0.050 Å2, respectively) compared with the other four atoms (0.0150.022 Å2), indicating that the former sites could be occupied by Au/In mixtures or pure In. Therefore, In4 and In5 were temporarily assigned to the former Au4 and Au5 atoms in subsequent refinements. At this stage, only In4 had an abnormal Ueq value (0.0089 Å2) compared with those of other atoms (0.0240.032 Å2), indicating a Au/In mixture at this site. This was so assigned in the final anisotropic refinements, in which R values drastically decreased to R1 = 5.84%, wR2 = 14.87% for 42 variables and 735 independent reflections. The refined composition is SrAu4.30(1)In1.70(1), consistent with the energy dispersive X-ray (EDX) data (SrAu4.1(1)In1.9(1)). The maximum and minimum peaks in the final difference Fourier map were 3.66 e/Å 3 (1.85 Å to Sr) and 4.04 e/Å3 (1.55 Å to Au2), respectively. Direct methods for 2 located 15 atoms, all with interatomic separations reasonable for Ag/InAg/In pairs (∼ 2.82.9 Å), so Ag1Ag15 were temporarily assigned to these sites. A few cycles of refinements resulted R1 = ∼ 19%, and the difference Fourier map yielded four sites apparently suitable for polar CaAg/In contacts (∼ 3.2 Å). They were so assigned to Ca1Ca4 in subsequent refinements, which converged at R1 = ∼12%. At this stage, the difference Fourier map indicated two additional sites (Ag16 and Ag17) with smaller peaks and shorter distances to neighboring atoms. Ag16 was observed to be equivalent to Cd16 in the defect-free, structurally closely related Na26Cd141,21 whereas Ag17 could be considered as part of a split site of Ag16 (below). Not surprisingly, both Ag16 and Ag17 had fractional occupancies and short distances to respective neighbors. Separate refinements to check the occupancies of Ag1Ag15 and Ca1Ca4 sites were also tried. The refined occupancy values for Ag1Ag15 and Ca1Ca3 were in the range of 0.98 (3)1.02 (3), indicating full occupancies for these sites, but the occupancy for Ca4 was refined to 0.90(4). Additional refinements
a
11092
dx.doi.org/10.1021/ic2016275 |Inorg. Chem. 2011, 50, 11091–11098
Inorganic Chemistry
ARTICLE
Table 2. Atomic Coordinates and Equivalent Isotropic Displacement Parameters for SrAu4.30(1)In1.70(1) (1)a atom
a
z
Ueq (Å2)
1
0.0857(2)
0.022(1)
0.4047(2)
1
0.0177(2)
0.022(1)
0.3158(2)
1
0.2637(2)
0.028(1)
0.0630(2)
0.5131(3)
0.3038(2)
0.026(1)
.m.
0.1447(4)
1
0.8603(3)
0.020(1)
.m.
0.2588(5)
1
0.5590(4)
0.019(1)
x
Wyck.
symm.
occ.
Au1
4c
.m.
0.0701(2)
Au2
4c
.m.
Au3
4c
.m.
Au/In
8d
1
In
4c
Sr
4c
0.65/0.35(2)
y /4 /4 /4 /4 /4
Ueq is defined as one third of the trace of the orthogonalized Uij tensor.
Table 3. Atomic Coordinates and Equivalent Isotropic Displacement Parameters for CaAg3.54(1)In1.88(1) (2)a occ. 6¼ 1
atomb
Wyck.
M1 M2
6l 12p
M3
12o
.m.
0.2608(1)
0.5216(1)
0.1558(1)
0.018(1)
M4
6j
m2m
0.4150(1)
0
0
0.022(1)
M5
12o
.m.
0.1803(1)
0.3607(1)
0.2515(1)
0.018(1)
M6
6j
M2m
0.1364(1)
0
0
0.032(1)
M7
12o
.m.
0.6161(1)
0.2323(1)
0.3338(1)
0.023(1)
M8
2c
6m2
1/3
2/3
0
0.024(1)
M9 M10
24r 6k
1 m2m
0.1095(1) 0.3884(1)
0.4545(1) 0
0.2747(1) 1/2
0.024(1) 0.025(1)
M11
6m
mm2
0.5503(1)
0.1005(1)
1/2
0.025(1)
M12
6i
2mm
1/2
0
0.2497(2)
0.021(1)
M13
12q
m..
0.0876(1)
0.3156(1)
1/2
0.042(1)
M14
12o
.m.
0.0935(1)
0.1869(1)
2602(1)
0.045(1)
M15
6k
m2m
0.1523(1)
0
1/2
0.042(1)
M16
1a
6/mmm
0.31 (1)
0
0
0
0.025c
M17 Ca1
2e 6l
6mm mm2
0.257(8)
0 0.5750(1)
0 0.1499(2)
0.091(1) 0
0.025c 0.020(1)
symm. mm2 m..
x
y
0.1354(1) 0.1317(1)
0.2708(1) 0.4081(1)
z 0 0
Ueq (Å2) 0.017(1) 0.016(1)
Ca2
12n
..m
0.2759(1)
0
0.2028(3)
0.021(1)
Ca3
6m
mm2
0.2393(1)
0.4786(2)
1/2
0.026(1)
Ca4
2e
6mm
0
0
0.278(3)
0.084(8)
0.90(4)
a
Ueq is defined as one third of the trace of the orthogonalized Uij tensor. b M = Ag, In, or Ag/In as they are indistinguishable in X-ray diffraction. c Fixed values in refinements. were also tried with all Ag sites being assigned as In, but no significant change was observed judging from R values, Ueq, or other related parameters (e.g., residual peak heights) within our imagination because Ag and In are close in atomic number Z. However, attempts to assign Ag/In admixtures to different sites resulted in divergent refinements. Apparently, Ag and In are indistinguishable in this X-ray solution. For the best approximation, Cd was used in the following leastsquares refinements because it has the average atomic number of Ag and In. (The scattering coefficients of Cd are actually not an average of Ag and In, though.) Nevertheless, M will be used to represent the indistinguishable Ag/In mixtures in later text. The final refinements yielded R1 = 3.75%, wR2 = 7.87% for 102 variables and 1713 independent reflections. The refined composition was Ca25.80(4)Cd140.82(3), a proportion very close to that of Na26Cd141. Since this was a pure phase product (Supporting Information, Figure S1), the formula for 2 is represented by the loaded ratio, Ca26.0(2)Ag92.0(1)In49.0(2) or, normalized, CaAg3.54(1) In1.88(1), which agrees well with the EDX data. As expected, the refinements of 2 could result in high residues in a difference Fourier map because of mismatches in X-ray scattering coefficients, which in turn might be mistakenly assigned to atoms with fractional occupancies. In this case, the maximum and minimum residual
peaks are 8.64 e/Å3 at (0 0 1/2) and 4.55 e/Å3 at (0 0 0.207) sites, respectively. However, no meaningful electron densities were found at the corresponding sites in the observed Fourier map (Supporting Information, Figure S2). Large residues in the difference Fourier map were also reported in the structure of Y13Pd40Sn31.22 Other arguable imperfections are the isotropic parameters for M6, M13M15, and Ca4; all show considerably larger Ueq values than the average of other atoms. The reason is because M6, M14, and M15 define a hexagonal tunnel, in which the disordered M16 and M17 alternate with Ca4 along the c axis (Supporting Information, Figure S2). M13 is also affected by the disorder phenomena as it is coplanar with M15 on the waist of the hexagonal tunnel. A summary of crystal and structural refinement data for both crystals is given in Table 1, and the refined positional parameters for 1 and 2 are given in Table 2 and Table 3, respectively, both standardized with TIDY.23 Other detailed crystallographic data are available in the cif file (Supporting Information). Electronic Structure Calculations. Electronic structure calculations were performed by the self-consistent, tight-binding, linear-muffin-tinorbital (LMTO) method2426 in the local density (LDA) and atomic sphere (ASA) approximations, within the framework of the density 11093
dx.doi.org/10.1021/ic2016275 |Inorg. Chem. 2011, 50, 11091–11098
Inorganic Chemistry
ARTICLE
functional theory (DFT) method.27 Interstitial spheres were introduced to achieve space filling. The ASA radii as well as the positions and radii of these empty spheres were calculated automatically under the limitation of 16%, 18%, and 20% for atomatom, atom-interstitial, and interstitial-interstitial sphere overlaps. Reciprocal space integrations were carried out using the tetrahedron method. The basis sets for Sr, Au, and In were 5s/(5p)/4d, 6s/ 6p/5d/(5f), 5s/5p/(5d); and for Na and Cd, 3s/(3p) and 5s/5p/4d/(4f). The orbitals in parentheses were down-folded in the calculations. Scalar relativistic effects were included in the calculations. The band structure was sampled for 8 8 8 k points for Sr3Au13In5 and 12 12 4 for Na26Cd141 in the irreducible wedge of the Brillouin zone.
electron count per atom (e/a) values fall within a range of 2.02.4. In contrast the present SrAu4.3In1.7 phase (red) lies in a lower e/a region, with e/a = 1.63. This e/a value is close to those of Ca3Au12.4In6.1 (normalized as CaAu4.1In2.0), the 1/1 AC (1.73) in the CaAuIn system14 and of Ca13Au57.1Ga23.4 (normalized as CaAu4.4In1.8), the 2/1 AC (1.64) in the CaAuGa system.13 The absence of QC/AC phases in SrAuIn system appears to have a relation with the larger size of Sr over Ca. Crystal 1 has a very different structure motif. This is manifested by the building blocks and their arrangements. All other structures exhibit various hexagonal (or/and pentagonal) tunnels derived from SrAuIn3 (BaAl4-type)18 except for the simplest SrAuIn.28 Table 4 summarizes the coordination numbers of Sr versus (Au + In)/Sr proportions, as well as Au (at.) %, and e/a for all ternary phases. Generally, high coordination numbers are found for compounds with larger (Au + In)/Sr proportions.31 This is because such large proportions mean that the electronegative components Au and In that define the networks around Sr need to be more condensed, resulting in the formation of more complex structures, as observed from SrAuIn28 to SrAu2In233 and to SrAu3In3.18 Crystal 1 has the largest coordination number (19) even though its (Au + In)/Sr proportion is not the largest. It seems that increases of Au atomic percentages, hence decreases of e/a, result in more compact structures with more delocalized and homogeneous covalent bonding. Structure of SrAu4.3In1.7. SrAu4.30(1)In1.70(1) (oP28), crystal 1, is isostructural with the parent binary CeCu6.34 Other ternary
’ RESULTS AND DISCUSSION Phases in SrAuIn System. Figure 1 shows the distribution of known ternary phases in the SrAuIn system; SrAuIn3 (I4mm),28 SrAu1.06In2.94 (I4/mmm),29 Sr2Au2.85In4.15 (P62m),30 Sr4Au9In13 (P6m2),31 SrAu3.35In4.65 (P21/m),32 SrAu3.76In4.24 (Pnma),31 SrAuIn (Pnma),28 SrAu2In2 (Pnma),33 and SrAu3In3 (Pmmn),18 which are numbered from 1 to 9 in the figure. As shown, these phases aggregate in a region in which the valence
Figure 1. Distribution of ternary phases in the SrAuIn system. Black spheres denote phases reported in literature, and the red sphere denotes the phase discovered in present work. The color gradient of background is scaled by the valence electron count per atom (e/a), as indicated at bottom.
Figure 2. 19-Vertex polyhedron centered by Sr in SrAu4.3In1.7. (M = Au/In).
Table 4. Relationship between Sr Coordination Numbers versus (Au+In)/Sr Ratios, Au (at.) %, and e/a in SrAuIn Ternary Phases phase
a
S. G.
(Au + In)/Sr
Au%
e/a
coordination numbersa
reference
SrAuIn
Pnma
2
0.33
2.00
12
SrAuIn3
I4mm
4
0.20
2.40
16
28
SrAu1.1In2.9
I4/mmm
4
0.21
2.38
16
29 30
28
Sr2Au2.8In4.2
P62m
3.5
0.32
2.14
16 + 15 + 13
Sr4Au9In13
P6m2
5.5
0.35
2.15
18, 16
31
SrAu3.3In4.7
P21/m
8
0.37
2.14
18
32
SrAu3.8In4.2
Pnma
8
0.42
2.05
18
31
SrAu2In2 SrAu3In3
Pnma Pmmn
4 6
0.40 0.43
2.00 2.00
16 18
33 18
SrAu4.3In1.7
Pnma
6
0.61
1.63
19
this work
The coordination numbers include only those with center-to-vertex distances 5.2 Å. The bond distances for AuAu, AuIn, AuM, InM, and MM are comparable with the sum of single bond metallic radii (Au, 1.339; In, 1.42139), except that dAu1In is shorter than expected, 2.682 (4) Å. A similar distance is found in Sr4Au9In13 (2.66 Å)31 and is explained by the smaller number of neighbors about each. The same reason applies to 1, see Table 5. The 19-vertex clusters are rarely seen in other structure types; rather, 18- or 20-vertex clusters are more common. For examples, 20-vertex dodecahedral clusters are often found in Zintl phases containing heavier triels (Ga, In, Tl).6 From the viewpoint of geometry, the Sr@Au9In4M6 cluster in 1 vividly mimics the 18vertex clusters in Ca3Au12.4In6.1 (YCd6-type),14 except that the latter contains two pentagonal caps and a hexagonal ring at the waist. Apparently, the increase of coordination number from Ca to Sr is required by size; the larger Sr evidently needs more anionic neighbors to achieve space filling. Each unit cell of SrAu4.3In1.7 consists of four Sr@Au9In4M6 clusters, two at y = 0.25 and two at y = 0.75, Figure 3. These polyhedra share the M2In triangle faces with each other in (101) and (011) directions. In the (010) direction, each polyhedron shares its quadrangular planes (Au1 f Au3 f Au2 f In) with other two polyhedra to form a polyhedral column along b, see Figure 3 b. The structure can also be considered as a pseudo layered structure along the b axis: “SrAu3In” layers at y = 1/4 and 3/4 plus M layers between. This is evident in the y coordinates in Table 2. However, the mixed Au/In form strong bonds with atoms from neighboring “SrAu3In” layers, particularly the bonding interactions
Figure 3. Packing of the 19-vertex polyhedral clusters in the unit cell of SrAu4.3In1.7 viewed along (a) [010] and (b) [100] directions.
with Au as suggested by their bond distances dMAu = 2.7762.918 Å, Table 5, diminishing the layered structural features. Structure of CaAg3.6In1.9. The structure of CaAg3.6In1.9, crystal 2, is related to both Na26Cd141 (hP167)21 and Y13Pd40Sn31 (hP168),22 which have the same space group (P6/mmm) and are free of any detectable defects. Both prototypes contain four independent sites for active metals (Na or Y) and 16 sites for electronegative components. Their differences lie mainly in two crystallographic sites: in Y13Pd40Sn31, the Wyckoff 1b (0 0 0) site is empty but the Wyckoff 2d (1/3 2/3 1/2) site is occupied by a Sn atom (Sn1). In contrast, a Cd atom (Cd16) is assigned to the 1b site with nothing at 2d in Na26Cd141. As a result, Y13Pd40Sn31 contains one more atom in the unit cell compared with Na26Cd141. The 11095
dx.doi.org/10.1021/ic2016275 |Inorg. Chem. 2011, 50, 11091–11098
Inorganic Chemistry
ARTICLE
Figure 4. Environments of Ca1Ca4 (ad) in CaAg3.5In1.9 (2), together with the structurally related (e) Y1-cluster in Y13Pd40Sn31 and (f) Na4-cluster in Na26Cd141. Red and green spheres in (ad) represent Ca and Ag/In atoms, as listed in Table 3. Violet, gray, red in (e) represent Pd, Sn, and Y, and cyan and red in (f), Cd and Na, respectively.
Figure 5. Polyhedral representations of the structure of CaAg3.5In1.9 projected in (a) (001) and (b) ∼(010) directions. Blue, yellow, red, and green represent Ca1, Ca2, Ca3, and Ca4-centered polyhedra, respectively.
structure of crystal 2 is closer to Na26Cd141, with the Wyckoff 1b sites occupied by M16 (M = Ag/In) but the 2d site unoccupied. However, the M16 site exhibits configurational disorder with M17 (below).
Figure 4 ad show the coordination polyhedra of Ca1Ca4, together with the polyhedra from Y13Pd40Sn31 (e) and Na26Cd141 (f). Each Ca1 has 17 neighbors that define pentagonal prisms with all faces capped. In comparison, both Ca2 and Ca3 have 16 neighbors, which define a pentagonal prism similar to those around Ca1 except that one rectangle face (1f1f2f2 for Ca2 and 7f7f7f7 for Ca3) on the waist is uncapped. Note that the 1f1f2f2 open rectangular face for Ca2 happens to be coplanar with two neighboring capping atoms (M4 and M6) to generate a hexagonal face. The last is shared with another Ca2 polyhedron. That is, two neighboring Ca2 polyhedra pair up to form a dimer, Figure 4 c, with the Ca2Ca2 distance about 3.33 Å. This reminds one of the similar Ca2 dimers in Ca14Au46Sn5,17 Ca13Au57.1Ga23.4,13 and Ca12.6Au37.0In39.6,14 all of which have short CaCa distances (